Objective Definition - The Forgotten Step Before a Big Decision
Updated: Aug 16
TL;DR / Key Takeaways
Core Concept: Decision-makers consistently mistake downstream actions or solutions ("I want to switch vendors") for actual objectives ("lower costs or fewer late shipments"). Pfleiderer’s Objective Identification Framework forces users to articulate what they are actually optimizing for before evaluating options.
The Zero-Suggestion Constraint: The system must never name, suggest, or frame objectives for the user. Self-articulated goals drive real optimization, whereas suggested goals are accepted politely and then quietly ignored during real tradeoffs.
Objectives vs. Constraints: "Keeping costs under $X$" is a boundary limit (constraint), not an outcome to maximize (objective). Conflating them prematurely shrinks the available solution space.
Protocol Mechanics: Uncovers root goals using multi-level "Why" chains, scans 6 dimensional blind spots, forces tier-ranking (must-haves vs. nice-to-haves), and stress-tests goals against non-catastrophic failure scenarios.
This article describes the core principles behind a Claude Skill that helps me with objective setting.
The principles behind the skill
Every decision process starts with the same silent shortcut: people name a solution and call it an objective. "I want to hire a new manager.", "I want to switch vendors.", "I want to go part-time.", “I want to start my own company.”. These are actually actions, guesses at what might get you somewhere, dressed up as the destination itself. Nobody wants to switch vendors for its own sake. They want lower cost, or fewer late shipments, or a supplier who actually picks up the phone. The action is downstream of something real and, for many, naming the something is hard.
This is the part almost everyone skips, because naming it feels slower than it is. Generating options feels like progress, while sitting with the question of what you're actually optimizing for feels like homework, and we don’t like homeworks.
When you delegate that homework to a chatbot, you get exactly what you asked for: a plausible, coherent-sounding answer built on top of whatever surface-level "objective" you typed in. The model has no mechanism to push back on a solution disguised as a goal, it just optimizes for the wrong target, fluently. A few grams of sycophancy dressed up as strategy. It sounds focused. It's useless, because the input was never interrogated.
The step people avoid because it's hard is usually the exact step that cannot be delegated.
Here's what actually surfacing an objective looks like when you don't skip it, and the discipline behind it is stricter than most people expect. The core rule: the person doing the coaching never names the objective for you.
"You NEVER name, suggest, or frame objectives for the user. You only help the user surface their own."The reasoning is simple. Objectives you articulate yourself, you'll genuinely optimize for. Objectives handed to you get nodded at and then quietly ignored the moment they conflict with what you actually care about. So the difference between a coaching question and a suggestion matters more than it looks:
❌ "It sounds like your real objective is to protect your reputation" this is a suggestion. ✅ "When you imagine this decision going badly, what specifically would be lost?", this is a question.First, an unfiltered inventory. Whatever you say you want, first pass, no editing. This is usually thin, and usually stated as an action rather than an outcome, which is the first thing to catch.
Then, why, more than once. Not once, most people stop one level too early. "Why does that matter to you? What does achieving that get you?" The first reason gets you to a real goal. The second one usually gets you to the actual one:
"Why does that matter to you?" "And what does that get you?" "If you achieved that, what would be better about your situation?" "Is there something even more fundamental underneath that?"Somewhere in this pass, objectives also get separated from constraints. "I need to keep costs under X" isn't something you're trying to achieve — it's a wall you can't walk through. Treating it as an objective shrinks your option space before you've even started evaluating anything.
Next, a scan for blind spots. Financial, relational, reputational, personal, temporal, organizational, most people arrive with one or two dimensions covered and the rest invisible. Not every decision needs all of them probed, but almost every decision has at least one gap in this list nobody thought to check.
Then, a forced ranking. A flat list of equally-weighted objectives isn't a list, it's a way of avoiding the tradeoff conversation until later, when it's more expensive. "If you had to rank them, which one, if you failed to achieve it, would make the whole decision feel like a failure?" People resist this step more than any other. It's also the one that makes the evaluation phase almost mechanical afterward.
Finally, a stress test before the list closes. Play the decision going badly, not catastrophically, just badly, and see what surfaces that weren't on the list yet. Check whether anyone else affected has objectives that aren't represented. Read the final set back and ask, one more time, whether each item is an outcome you're chasing or still an action wearing an outcome's clothes.
The failure patterns repeat enough that they're worth naming on sight:
Pattern | Signal |
Solution-as-objective | "My goal is to hire / buy / switch / implement X" |
Shallow why | Stops at the first-level objective without probing deeper |
Missing stakeholders | Only your own objectives get named |
Constraint mistaken for objective | "I need to keep costs under X" |
Tension blindness | Two objectives quietly pulling in opposite directions |
Premature completeness | "That's all of them", before the gap scan even happens |
None of this is complicated. It's slow, and it requires something that won't let you skip it, which is precisely what an AI optimized for a fast, confident-sounding answer will never do.
Once the objective set is genuinely clear, traced through "why," checked against blind spots, weighed against itself, everything downstream moves faster. Options stop feeling infinite. Evaluation stops requiring a fresh debate every time, because you already know what you're weighing against what. Decisions stop needing justification after the fact, because they were built on the right foundation from the start.
I built a skill for myself that runs this process end to end. I don't use it for what to have for dinner, but rather every time a decision actually matters.
The anatomy of the skill
Metadata & Trigger Conditions
Trigger Expressions: Activates on phrases like "I need to decide whether to...", "my goal is to...", "help me think through what I'm trying to achieve", or whenever options are brought up before underlying goals are explicit.
System Integration: Operates standalone or serves as Stage 1 of the broader Decision-Making Operating System before transitioning into alternatives expansion.
The Absolute Socratic Constraint
Non-Negotiable Guardrail: Forbids suggesting goals, validating inputs with praise ("great point"), or phrasing insights as suggestions.
"You NEVER name, suggest, or frame objectives for the user. You only help the user surface their own." Allowed Coaching Vector: Cross-domain examples are permitted solely to demonstrate logic, provided they close with a question returning the thinking to the user.
The 5-Step Identification Protocol
Technique 1 — Surface Starting Objectives
Collects an unfiltered initial inventory and instantly flags actions wearing an outcome's clothes.
"That sounds like an action — what would that action get you? What's the outcome you're actually after?"
Technique 2 — Multi-Level "Why" Chain & Constraint Separation
Probes 2–3 levels deep ("Why does that matter? What does achieving that get you?") until hitting fundamental root goals.
Differentiates goals to maximize (objectives) from lines that cannot be crossed (constraints).
Technique 3 — 6-Dimensional Blind Spot Scan
Systematically checks unmentioned dimensions for hidden gaps: Financial, Relational, Reputational, Personal, Temporal, and Organizational.
Technique 4 — Forced Prioritization & Tension Detection
Breaks flat lists by forcing tier rankings (must-haves vs. nice-to-haves) and explicitly surfacing conflicting goals.
"If you had to rank them — which one, if you failed to achieve it, would make the whole decision feel like a failure?"
Technique 5 — Multi-Angle Stress-Testing
Audits the finalized set against non-catastrophic failure scenarios, unrepresented stakeholder needs, and leftover disguised solutions.
Objective Clarity Watch & Pattern Recognition
Active Behavioral Audit: Continuously monitors for six core failure modes: Solution-as-Objective, Shallow Why, Missing Stakeholders, Constraint as Objective, Tension Blindness, and Premature Completeness.
System Synthesis & Programmatic Handoff
Output Artifact: Groups confirmed outcomes by priority tiers, separates constraints, and flags active tradeoffs.
Next Stage Transition: Automatically routes the explicit objective baseline into Stage 2 (Alternatives Expansion).
"You now have a clear, prioritized set of objectives — not just a surface-level goal. The next step is to make sure the options you're considering are actually wide enough to give these objectives a fair chance."
Want to put this + and the rest of the decision-making skills to work on something you're actually facing? I am giving away the skills for an one-hour discovery interview.
Below a friend of mine trying the objective definition skill - a few screens, only. The conv ersation went much deeper.
